Google ADK Research Agent

A powerful AI research assistant that performs multi-step web research with reflection and iterative improvement to provide thorough, well-cited answers to your questions.

✨ Features

• Intelligent Research: Automatically generates diverse search queries from your questions
• Comprehensive Analysis: Performs multiple web searches and synthesizes findings
• Quality Assurance: Analyzes research quality and conducts follow-up searches if needed
• Real-time Updates: Live progress tracking as research unfolds
• Flexible Depth: Choose between quick, thorough, or comprehensive research
• Proper Citations: All answers include sources and citations

🚀 Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+
• Google API Key with Gemini access

Setup

1. Clone and install

git clone <repository-url>
cd research-google-agent-adk

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # Linux/Mac
# or .venv\Scripts\activate  # Windows

# Install dependencies
pip install -r adk-backend/requirements.txt

2. Configure API key

# Copy example environment file
cp .env.example .env

# Edit .env file with your Google API key
nano .env

3. Run the application

# Start both frontend and backend
./run_full_stack.sh

4. Access the application

• Open http://localhost:5173 in your browser
• Backend API available at http://localhost:8000

🔍 How to Use

1. Enter your question - Type any research question
2. Select effort level:
   • Low: Quick research (1 search, 1 iteration)
   • Medium: Thorough research (3 searches, up to 3 iterations)
   • High: Comprehensive research (5 searches, up to 10 iterations)
3. Choose AI model - Select between available Gemini models
4. Submit - Watch the research progress in real-time
5. Get results - Receive a comprehensive answer with citations

🛠️ Configuration

Environment Variables

Create a .env file with:

GOOGLE_API_KEY=your-google-api-key-here
GOOGLE_GENAI_USE_VERTEXAI=FALSE

Available Models

• gemini-2.0-flash - Fast, efficient research
• gemini-2.5-flash-preview-04-17 - Latest flash model (default)
• gemini-2.5-pro-preview-05-06 - Highest quality responses

📡 API Reference

Stream Research Results

POST /api/adk-research-stream

Request:

{
  "question": "What are the latest advances in quantum computing?",
  "effort_level": "medium",
  "reasoning_model": "gemini-2.5-flash-preview-04-17"
}

Response: Server-Sent Events stream with real-time updates

🏗️ Architecture

The system uses Google's Agent Development Kit (ADK) with an enhanced orchestrator that manages:

• Query generation from user questions
• Parallel web searches using Google Search API
• Quality analysis and gap identification
• Iterative research loops for comprehensive coverage
• Final answer synthesis with proper citations

📄 License

MIT License - see LICENSE file for details